Cytotoxic chemotherapy produces modest benefits for patients with recurrent and metastatic squamous cell carcinoma of the head and neck (SCCHN). Prospective randomized clinical trials have failed to demonstrate unequivocal superiority of aggressive multidrug regimens over single agents. Despite this, phase II trials frequently result in encouraging preliminary observations that compare favorably to historical single-agent data. While providing for a useful method of screening for anti-tumor activity, phase II studies have limited use in determining the relative value of a new treatment program. Results of phase II studies are considerably influenced by patient selection factors and criteria used to establish therapeutic benefits (responses). Furthermore, estimations of true levels of efficacy (response rates) are dependent on sample sizes, which are usually limited in such trials. We propose that newly developed combinations containing at least one known active agent in this disease should be tested in a controlled setting after their toxicity pattern has been well established. The conduct of the usual phase II study in these situations will probably not provide useful new information, since responses are likely to be observed. We describe a two-stage design applied to terminate a trial if at the first stage there is no evidence of improvement over the control arm. This method allows for early termination of studies involving relatively inefficient treatment regimens and, at the same time, continuation of those with a high likelihood to result in significant therapeutic improvements over a control arm. Loss of power is negligible and sample sizes can be reduced significantly. The rationale behind this method and its simplicity are attractive features for a widespread application for new drug development strategies in this and other diseases.